Sonlicromanol’s active metabolite KH176m normalizes prostate cancer stem cell mPGES-1 overexpression and inhibits cancer spheroid growth

Aggressiveness of cancers, like prostate cancer, has been found to be associated with elevated expression of the microsomal prostaglandin E synthase-1 (mPGES-1). Here, we investigated whether KH176m (the active metabolite of sonlicromanol), a recently discovered selective mPGES-1 inhibitor, could affect prostate cancer cells-derived spheroid growth. We demonstrated that KH176m suppressed mPGES-1 expression and growth of DU145 (high mPGES-1 expression)-derived spheroids, while it had no effect on the LNCaP cell line, which has low mPGES-1 expression. By addition of exogenous PGE2, we found that the effect of KH176m on mPGES-1 expression and spheroid growth is due to the inhibition of a PGE2-driven positive feedback control-loop of mPGES-1 transcriptional regulation. Cancer stem cells (CSCs) are a subset of cancer cells exhibiting the ability of self-renewal, plasticity, and initiating and maintaining tumor growth. Our data shows that mPGES-1 is specifically expressed in this CSCs subpopulation (CD44+CD24-). KH176m inhibited the expression of mPGES-1 and reduced the growth of spheroids derived from the CSC. Based on the results obtained we propose selective mPGES-1 targeting by the sonlicromanol metabolite KH176m as a potential novel treatment approach for cancer patients with high mPGES-1 expression.


Introduction
Prostate cancer (PCa) is the most frequently diagnosed cancer in the Western world [1]. It is also the leading cause of cancer-related death in males over 65 years of age [2]. Currently, PCa is treated with androgen deprivation and chemotherapeutic agents, but there is an unmet medical need for novel drug targets considering the relatively poor outcome of current therapeutic interventions. The microsomal prostaglandin E synthase 1 (mPGES-1) has been found overexpressed in PCa (48% in organ-confined PCa and 77.7% advanced PCa [3] and might therefore be such a target as has been suggested [4].

Cultures of spheroids in Matrigel
Cells were detached with trypsin and either sorted or unsorted cells (140000) were resuspended in 50 μL culture medium and gently mixed with 1 mL Matrigel matrix. 50 μL (7000 cells) of the mixture was placed as a drop into a 24-well-plate, and incubated at 37˚C for 20 min. Then 1 mL of warm culture medium was added on top of the gel. After 24 h, 700 μL medium of each well was replaced by fresh medium containing KH176m or vehicle (0.1% DMSO) in presence or absence of exogenous PGE 2 (1-100 nM) to reach the indicated final concentrations. This medium was refreshed every 2 days by gently discarding 700 μL medium of each well and replacing with the same volume fresh warm medium containing KH176m or vehicle (0.1% DMSO) in presence or absence of exogenous PGE2 (1-100 nM) to reach the indicated final concentrations. After 7 days of treatment, spheroids were used for different subsequent analysis.

Cultures of single spheroid in 96-well ultra-low attachment plate
The suspension of sorted cells was diluted with complete culture medium to obtain the final density (2500 cells/mL) and dispensed (200 μL/well) into 96-well ultra-low attachment plate (Corning, New York, USA). The plate was centrifuged at 300 g for 5 min and placed in an incubator (5% CO 2 , 37˚C). After 24 h, 100 μL medium of each well was replaced by fresh medium containing 6 μM KH176m or vehicle (0.2% DMSO) (final concentration is 3 μM KH176m or 0.1% DMSO). This medium was refreshed every 2 days by gently discarding 100 μL medium of each well and replacing with the same volume fresh warm medium containing 3 μM KH176m or vehicle (0.1% DMSO) to reach the indicated final concentrations. After 7 days of treatment, spheroids were analyzed by immunofluorescence.
spheroids were harvested by gently pipetting and washed in cold PBS twice. Spheroids were disintegrated by treatment with 2 mL Trypel (Gibco, Landsmeer, the Netherlands) containing 2 μL DNAse I (Sigma-Aldrich, Zwijndrecht, the Netherlands) per sample (incubated at 37˚C for 20 min) with gently pipetting of the suspension every 5 min. Single cells were then stained with Fixable Viability Dyes (FVD) eFluorTM 450 (eBioscience, Landsmeer, the Netherlands) for 20 min on ice. After washing with cold PBA (PBS containing 0.5% BSA and 0.01% NaN3), cells were blocked by 2% human serum (Sigma-Aldrich, Zwijndrecht, the Netherlands) for 15 min on ice. Following incubation with anti-CD24-APC (eBioscience, Landsmeer, the Netherlands) and anti-CD44-PE (eBioscience, Landsmeer, the Netherlands), for 30 min on ice, cells were washed twice and re-suspended with cold PBA. Flow cytometry analysis was carried out on a BD FACS Verse (BD Biosciences) and fluorescence-activated cell sorting (FACS) experiments on a BD FACS ARIA III (BD Biosciences) instrument. Subsequent flow cytometry data were analyzed in FlowJo V10 software (Tree Star).

Analysis of spheroid number and size by immunofluorescence
For spheroids culture in Matrigel, the medium from spheroids was carefully removed, and 500 μL/well 30 μM Calcein-AM Viability Dye (Invitrogen, Landsmeer, the Netherlands) which was prepared in warm culture medium was added and incubated at 37˚C for 30 min after which 500 μL/well of warm culture medium was added to each well. For single spheroid cultured in 96-well ultra-low attachment plate, 100 μL medium was replaced by the same volume of medium containing 10 μM Calcein-AM Viability Dye (Invitrogen, Landsmeer, the Netherlands) and incubated at 37˚C for 20 min. Fluorescent images were acquired in confocal Z stack mode using a BD Pathway 1 855 system, 200 sections (5 μM / section), which covered the entire thickness of the sample, were collapsed into a single 2D image for further analysis with ImageJ Pro. For spheroids cultured in Matrigel, total cell area of each image indicated by Calcein-AM positive staining was obtained by using ImageJ Pro software. After blinding images, the number of spheroids was counted manually. For single spheroid, the area of each spheroid was counted manually by using ImageJ Pro software after blinding of the images. Results were expressed as average pixels of each spheroid (total area of spheroids / number of spheroids) and data was normalized to vehicle (%).

RNA extraction and qRT-PCR
Total RNA was isolated from spheroids by using the TRIzol reagent (Invitrogen, Uden, the Netherlands). The obtained mRNA was reverse transcribed to cDNA from 2 μg of total RNA using a FirstStrand cDNA Synthesis Kit (Roche, Woerden, the Netherlands). Quantitative PCR analysis was performed in a total volume of 20 μL containing cDNA template, sense and antisense primers, and SYBR 1 Green master mix (QIAGEN, Venlo, the Netherlands). Data was expressed as fold changes relative to vehicle after normalization to the housekeeping gene GAPDH using the ΔΔ CT method [37]. Each PCR was performed in duplicate on two separate occasions from at least three independent experiments (primer information is shown in S1 Table).

Western blot analysis
The medium from spheroids was carefully removed, and 1 mL/well ice-cold PBS was added and incubated on ice for 5 min. Then spheroids were harvested by gently pipetting of the suspension, and then spheroids were washed with ice-cold PBS for three times. Cells from conventional 2D cultures were collecting by trypsinization and then washed with ice-cold PBS twice. Spheroids or cells were lysed in buffer (50 mM Tris-HCl pH8.0, 150 mM NaCl, 0.2%

Statistical analysis
All experiments were independently performed in triplicate, and the results were presented as mean ± S.D. Statistical analysis was performed with GraphPad Prism (GraphPad Prism 7.0 Software). Experiments were designed to determine whether the effects of treatment were dependent on vehicle conditions. Variance between the experimental groups was determined by Student t-test. P<0.05 was considered statistically significant. Information about the number of samples (n) is included in the figure legends.

Constitutive expression of mPGES-1 in DU145 PCa cells is reduced by KH176m
Elevated mPGES-1 is considered as an important factor in determining tumorigenic potential in PCa cells [3,38]. We have recently found that KH176m can reduce the expression of inflammation induced mPGES-1 levels [35]. To investigate the effect of KH176m on cells with high constitutive levels of mPGES-1 and to establish a cell model to study functional consequences of modulating PGE 2 levels, we employed human PCa cell lines with different levels of mPGES-1. In line with the work by Hanaka et al. [5], we could confirm that the expression of mPGES-1 is much higher in the PCa cell line DU145 as compared to LNCaP ( Fig 1A). Upon treatment with KH176m for 24 hours the constitutive expression of mPGES-1 in DU145 cells could be reduced in a dose-dependent manner ( Fig 1B). In order to investigate the effect of KH176m on the proliferation of these PCa cell lines, the cell proliferation was determined during 96 hours in the presence of various doses of KH176m. We did, however, not observe significant differences among the vehicle and KH176m treated groups (S1 Fig). Taken together, KH176m can block constitutive expression of mPGES-1 without affecting cell growth.

KH176m affects spheroid growth and expression of mPGES-1 of DU145 cells
To better mimic the in vivo environment of the tumor cells, we established a three-dimensional (3D) culture system that provides a more physiological relevant environment for cells since it supports processes such as cell-cell and cell-extracellular matrix (ECM) interactions [39]. To investigate the effect of KH176m on spheroid growth, DU145 and LNCaP cells were grown in Matrigel matrix which allows the formation of spheroid structures. These 3D cultures were grown for 7 days in the presence or absence of KH176m treatment. The average size of spheroids derived from DU145 cells was 3 times larger than those derived from LNCaP cells (Fig 2A and 2B). After treatment with KH176m for 7 days, the size of the spheroids derived from DU145 cells was significantly decreased. However, no changes were observed in spheroids derived from the LNCaP cells (Fig 2A and 2B). We hypothesized that the phenotypical changes that only occurred in the DU145 spheroids might be related to their high constitutive expression of mPGES-1. We therefore measured protein and mRNA levels of mPGES-1 and mRNA levels of other enzymes involved in prostaglandin synthesis in both DU145 and LNCaP derived spheroids. In line with the results using 2D cultured cells, mPGES-1 was highly expressed only in DU145 cells, and its protein and mRNA levels were decreased by KH176m treatment in a dose-dependent manner (Fig 2C and 2D). In LNCaP spheroids the mPGES-1 protein and mRNA levels were below the level of detection. In spheroids derived from both cell lines, the mRNA level of the upstream Cyclooxygenase 2 (COX-2) was too low to be detected, which may explain why PGE 2 levels in DU145 as well as LNCaP (both in 2D and 3D cultures) were below the level of detection, despite the high expression of mPGES-1 in the DU145 spheroids. Additionally, other constitutive genes involved in prostaglandin synthesis, including mPGES-2, cytosolic PGES (cPGES), and COX-1, did not show significant differences between the two cell lines and also not upon treatment (S2 Fig). In conclusion, the high level of mPGES-1 in DU145 cells correlated with its strong ability to form spheroids in 3D cultures. Treatment of these cultures with KH176m specifically reduced mPGES-1 levels and also reduced spheroid size.

Effect of KH176m on spheroid growth and expression of mPGES-1 is overcome by exogenous addition of PGE2
Based on our previous work in fibroblasts and macrophage-like RAW264.7 cells, addition of exogenous PGE 2 reversed the effect of KH176m on mPGES-1 expression, suggesting a PGE 2 -driven positive feedback control of mPGES-1 transcriptional regulation, which was directly inhibited by KH176m [35]. We therefore hypothesized that also in the PCa cells inhibition of mPGES-1 expression and spheroid growth by KH176m will be restored by administration of exogenous PGE 2 . We thus treated DU145 3D-spheroids with increasing concentrations of PGE 2 (1-100 nM) with or without KH176m for 7 days and measured spheroid growth. Indeed, we found that the decrease in spheroid size caused by KH176m treatment could be restored by PGE 2 addition in a dose-dependent manner. Addition of PGE 2 in the absence of KH176m had no effect on the spheroid growth (Fig 3A and 3B).
We measured the expression of mPGES-1 in the same experimental setup and found that mPGES-1 expression, which was inhibited by KH176m, was also restored by exogenous PGE 2 administration in a dose-dependent manner (Fig 3C). In the absence of KH176m, adding more than 1 nM of PGE 2 increased the mPGES-1 levels whereas under these conditions no increase of the spheroid size was observed. These results demonstrated that exogenous PGE 2 treatment reversed the effect of KH176m in DU145 spheroids, suggesting both mPGES-1 and PGE 2 are involved spheroid growth.

KH176m selectively decreases prostate cancer stem cell population
Recent evidence supports the model that the cancer stem cells (CSCs) are responsible for tumor initiation and formation [19]. As reported, the CD44 + CD24subpopulation of PCa cells are stem-like cells that are responsible for colony and tumor initiation [22]. To determine whether inhibition of mPGES-1 by KH176m affects the equilibrium between prostate CSCs and non-CSCs, we treated DU145 and LNCaP derived spheroids with KH176m and evaluated the proportion of each subpopulation. CSCs and non-CSCs subpopulations were counted by flow cytometry using the cancer stem cell markers CD44 and CD24 (Fig 4 and S3 Fig). Our data showed that the CD44 + CD24subpopulation (CSCs) differed considerably between the two studied cell lines. A higher content of CSCs was found in the DU145 cells, whereas only a small fraction of CSCs was present in LNCaP cells which might explain why DU145 line grow larger spheroids. Upon treatment with KH176m the fraction of CSCs in DU145 spheroids was significantly decreased in a dose-dependent manner (Fig 4A and 4B).

KH176m selectively inhibits mPGES-1 expression and affects spheroid growth in prostate cancer stem cells
To further investigate the effect of KH176m on prostate CSCs, we purified a population of CD44 + CD24cells from DU145 cells using flow cytometry. The purified CSCs and also the remaining non-CSCs were grown in Matrigel to allow the formation of spheroid structures. These 3D cultures were grown in the presence or absence of KH176m treatment for 7 days. Interestingly, the CD44 + CD24 -CSCs generated on average approximately 6 times larger spheroids than the non-CSCs (Fig 5A and 5B). These results indicate that the CSCs represent a near homogeneous population with respect to spheroid-initiating ability. Treatment with KH176m decreased spheroid size in the CSCs, whereas the small non-CSCs spheroids remained unchanged (Fig 5A and 5B). This was further confirmed by experiments using FACS-sorted cells CSC plated in ultra-low attachment plates in which cancer stem cells are growing in an undifferentiated state. After treated with 3 μM KH176m for 7 days, we only observed decreased size of spheroid in the CSCs population (Fig 5C). We also investigated whether KH176m could inhibit the proliferation of isolated CSCs as well as non-CSCs. Our results show that there is no significant cell growth inhibition up to the concentration (3uM of KH176m) that decrease cancer stem cell population as well as affect CSC-derived spheroids growth in both populations (S4 Fig).
To determine changes in gene expression underlying the phenotypic change induced by KH176m, qRT-PCR was performed on isolated CSCs and non-CSCs. Interestingly, the data showed that mPGES-1 was specifically expressed in CSCs but not in non-CSCs (Fig 5D). Moreover, the mPGES-1 mRNA level in CSCs was significantly decreased by KH176m in a dose dependent manner (Fig 5C). The mRNA levels of the other tested genes (mPGES-2, cPGES, and COX-1) remained unchanged in both CSCs and non-CSCs after treatment with KH176m ( S5 Fig). Both of these results are consistent with our previous observations in the unsorted DU145 cells which show constitutive expression of mPGES-1. The KH176m-induced decrease in size of CSCs-derived spheroids is consistent with the reduction in mPGES-1 mRNA level, indicating that mPGES-1 might be a key regulator in the maintenance of stem cells capacity. Together, these results show KH176m can influence the cancer stem cell equilibrium by inhibiting the mPGES-1 level.

Discussion
The present study shows that KH176m, a recently discovered selective mPGES-1 inhibitor, has the potential to decrease PCa tumor aggressiveness by inhibiting mPGES-1 expression. In a 3D PCa spheroid culture system we show that KH176m inhibits expression of mPGES-1 and growth of PCa cell-derived spheroids by influencing the CSCs population equilibrium.
As an experimental paradigm we employed DU145 and LNCaP cells in which DU145 cells has higher levels of mPGES-1 as compared to LNCaP ( [5], and this study). Recently, our lab demonstrated that the sonlicromanol metabolite, KH176m, selectively inhibits mPGES-1.
Here, we undertook experiments to test whether the compound could modify prostate tumor formation. We showed that KH176m could reduce the constitutive expression of mPGES-1 in DU145, however without affecting the growth rate of this cell line in conventional 2D cultures.
To better understand the effect of KH176m on prostate tumorigenesis, we next employed a 3D spheroid in vitro culture model that better mimics the in vivo environment and provides a more accurate drug response as compared to traditional 2D cell cultures [40]. Our data showed that, whereas both PCa cell lines can form spheroids in Matrigel, the DU145 derived spheroids grow to a larger size than those from LNCaP cells. Base on literature findings, we hypothesized that the ability of spheroid formation was possibly linked to the different expression levels of mPGES-1 [3,5]. Indeed, we found that the mPGES-1 protein and mRNA expression were significantly decreased by KH176m in DU145 derived spheroids while other key genes involved in PGE 2 synthesis remained unchanged. Our recent study in fibroblasts and macrophage-like RAW264.7 cells has revealed that the effect of KH176m on mPGES-1 expression is due to the inhibition of a PGE 2 -driven positive feedback controlloop of mPGES-1 transcriptional [35]. An effect of KH176m on PGE 2 production could, however, not be determined in the spheroid cultures because the amount of PGE 2 that was secreted during their growth was below the level of detection, despite the high level of mPGES-1 present in DU145 spheroids. This may be due to low COX-2 expression and therefore limited availability of the mPGES-1 substrate PGH 2 . However, by adding exogenous PGE 2 to the DU145 derived spheroids we could counteract the effect of KH176m on spheroid growth and mPGES-1 expression. These findings suggest that both PGE 2 and mPGES-1 play a role in the oncogenic drive (Fig 6).
Studies have shown that increased levels of mPGES-1 correlate with a poor prognosis in PCa, suggesting that mPGES-1 may play a key role during PCa progression [22]. Selective inhibition of mPGES-1 is anticipated as a new strategy for anti-cancer treatment [41]. Also in cultured DU145 cells, the mPGES-1 expression level was found to correlate with tumorgenicity [3,5]. Previous studies also showed that knockdown or inhibition mPGES-1 in DU145 cells prevents the development of a vigorous tumorigenic phenotype, and affects stem-cell-like features (lower expression of CD44 and higher expression of CD24) [3]. In addition, CSCs are abundant in DU145 cells but not in LNCaP cells, leading to a greater clonogenic and tumorigenic properties of DU145 than LNCaP [22]. As CSCs are thought to be the tumor-initiating cells, and mPGES-1 expression in prostate cancer cells was clearly associated with stem-like features, we hypothesized that KH176m could influence the ratio between CSCs and non-CSCs cells, the CSCs equilibrium, by inhibiting mPGES-1 expression. Indeed, in spheroids from the DU145 cells, the fraction of CD44 + CD24marked CSCs was significantly decreased in presence of KH176m treatment. There are two hypotheses regarding the observed reduction of the CSCs fraction: (1) the CSCs differentiated into non-CSCs in presence of KH176m; (2) the growth of CSCs was blocked by KH176. Our results so far cannot differentiate between these two scenarios. Our data also showed that the purified CSCs from the DU145 cell line had a greater ability to form spheroids than non-CSCs. Furthermore, the ability of CSCs to form spheroids was inhibited by KH176m in a dose-dependent manner.
Drug resistant of CSCs is one of the limitations of conventional chemotherapy [42,43]. It is therefore urgent to develop novel therapeutic strategies that combine conventional chemotherapy with CSCs inhibitors [44][45][46][47]. It is important to note, however, that there are currently no known universal markers for CSCs that can be used for all tumor types, limiting the development of a CSCs targeting therapy for all patients [43]. Due to the complexity and diversity among CSCs, it is important to identify CSCs-specific markers to enable the development of customized therapies. Therefore, our finding that mPGES-1 was found to be specifically expressed in PCa CSCs is of particular interest.
In conclusion, our findings show that KH176m selectively inhibits mPGES-1 expression in the PCa CSCs population resulting in reduced spheroid growth. This may be of relevance for the treatment of patients with high expression of mPGES-1 and generally poor outcome. Furthermore, CSCs are thought to be the tumor-initiating cells, suggesting that the sonlicromanol metabolite KH176m could be considered as an anti-tumor drug based on its ability to decease spheroids size formed by cancer stem cells.